Ahrefs

Ahrefs' free tools gave us numbers we could check, and the numbers checked out. Every backlink we verified was real, every broken link it flagged returned the 404 it reported, and the Domain Rating held steady across two tools. The catch is dosage: the free version shows twenty rows where the paid product shows thousands, buckets keyword volumes into floors like over 1,000, and gates keyword difficulty behind a sign-up on half the list. Our limited 4/5 covers five automated anonymous runs of the free SEO tools, not the paid Ahrefs product.

What we tested. Q1 ran the Backlink Checker on python.org. Q2 ran the Website Authority Checker on the same domain. Q3 asked the Keyword Generator for ideas around ai tools in the United States. Q4 ran the Broken Link Checker on python.org. Q5 asked the Keyword Difficulty Checker about best ai tools. Each tool was submitted once with the intended input, with no account and no sign-in.

The backlink data. The Backlink Checker returned Domain Rating 92, 42M backlinks at 95% dofollow, 128K linking websites at 88% dofollow, and a twenty-row sample of referring pages with anchors and target URLs. We fetched the first three referring pages ourselves. All three contained a link to the exact target URL Ahrefs listed: online-python.com to www.python.org, w3schools.com/python to python.org, imagetotext.io to docs.python.org. That is the check that matters for a competitor backlink lookup, and it passed cleanly.

The authority check. The Website Authority Checker gave the same Domain Rating, 92, and the same linking-website count, 128K at 88% dofollow. It also reported 147M backlinks at 98% dofollow. That is not the 42M the Backlink Checker showed for the same domain minutes earlier. The Backlink Checker's view was labelled one link per domain, but that label cannot explain a 42M figure against 128K linking websites, and neither page says which count is which. Treat the DR and the linking-website count as the stable figures; treat the raw backlink total as a number whose definition varies by tool.

The keyword ideas. The Keyword Generator returned twenty ideas, every one containing ai tools, with a KD bucket on the first ten (Hard, Medium, Easy or N/A) and Sign up or a blank cell on the rest. Volume appeared only as a floor: over 10,000 for the seed, over 1,000 for everything else. Two of the twenty ideas were branded or junk strings (webnewstips com free ai tools, droven.io ai tools 2026), and two more read like scraped headlines. Useful for a first pass at a topic, not for prioritising a content calendar.

The broken links. The Broken Link Checker reported 29K broken links on python.org and 94K broken links pointing to it, with a five-row sample of outbound breaks (four distinct pages; one wiki page appeared under two URL variants). We fetched the first three flagged URLs: all three returned 404, as listed. The free view opens on outbound breaks; the inbound tab was not opened.

The difficulty score. The Keyword Difficulty Checker gave best ai tools a KD of 48, labelled Hard, with the estimate that followed backlinks from about 77 websites would be needed to rank in the top ten, and a ten-result SERP overview with DR, UR, backlinks, domains, traffic and keyword counts on the first five results and a sign-up gate on the rest. The Hard label agrees with the bucket the Keyword Generator showed for the same phrase.

Two observations outside the rating. Every result page carried an upsell: Need more? Try Starter plan, $29/mo, a Chrome toolbar prompt, and a free Ahrefs account nudge. None of them blocked a result. Our tooling also misfired twice before the difficulty check (a click that never landed, then a click on the example chip instead of the submit button); those were our errors, disclosed here, and the product was submitted once with the intended input.

Where this leaves the reader. If you need one competitor's DR, a sample of who links to them, or a list of dead links to chase, the free tools answer accurately and for nothing. If you need volumes as numbers, difficulty across a whole list, or more than twenty rows, you are looking at the paid product, which this review did not test. The verified accuracy and the cross-tool DR agreement are why this scope earns 4/5 rather than 3/5; the unexplained backlink-total gap and the gated, bucketed keyword data are why it does not earn 5/5. That is an editorial judgment, not a pass count.

What this review cannot establish. Testing was automated on September 14, 2026 in a Chromium browser at a 1280 x 1280 viewport, anonymously. We did not test Site Explorer, Keywords Explorer, Content Explorer, Site Audit, Rank Tracker, Ahrefs Webmaster Tools, the API, exports, AI features, index freshness claims, or the accuracy of volume and difficulty estimates against any external source. Sample sizes were the free caps (20 backlinks, 20 keywords, 5 referring pages, 10 SERP rows). These gaps bound the review rather than lower the score, and no product-wide rating follows from them.